[
https://issues.apache.org/jira/browse/PIG-3836?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13948593#comment-13948593
]
Cheolsoo Park commented on PIG-3836:
------------------------------------
[~amatsukawa], thank you for the patch. But it breaks the compilation.
{code}
[javac]
/Users/cheolsoop/workspace/pig-apache/src/org/apache/pig/newplan/logical/relational/LogicalPlan.java:130:
cannot find symbol
[javac] symbol : method murmur_32(int)
[javac] location: class com.google.common.hash.Hashing
[javac] HashFunction hf = Hashing.murmur_32(0);
[javac] ^
{code}
Did you mean {{murmur3_32(int)}}? I see it in JavaDoc
[here|http://docs.guava-libraries.googlecode.com/git-history/v11.0/javadoc/com/google/common/hash/Hashing.html#murmur3_32(int)].
> Pig signature has has guava version dependency
> ----------------------------------------------
>
> Key: PIG-3836
> URL: https://issues.apache.org/jira/browse/PIG-3836
> Project: Pig
> Issue Type: Bug
> Reporter: Akihiro Matsukawa
> Assignee: Akihiro Matsukawa
> Attachments: 0001-use-murmur-hash-with-0-seed.patch
>
>
> We currently use Hashing.goodFastHash(32) to generate signatures for the
> logical plan. Under the hood, this is a murmur32 hash.
> Guava 11, which pig directly depends on, always seeds the hash with 0.
> http://docs.guava-libraries.googlecode.com/git-history/v11.0/javadoc/src-html/com/google/common/hash/Hashing.html#line.85
> In future versions of Guava, it is seeded by current time:
> http://docs.guava-libraries.googlecode.com/git-history/v14.0/javadoc/src-html/com/google/common/hash/Hashing.html#line.47
> So when future versions of guava is pulled into the classpath (which if
> often), we no longer get the same signature for the same logical plan. This
> introduces unnecessary complexity for managing the classpath.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
